Proton T1r dispersion imaging holds great promise as a method of measuring and imaging 17O concentrations. However, careful choice of pulse sequence parameters is crucial to its proper operation. In particular, the choices of spin-locking times and spin-locking powers are of utmost importance in maximizing the sensitivity and specificity for 17O. We have begun to use a "propagation of error" analysis to determine the optimum spin-locking powers and times for tissues of varying relaxation properties and 17O concentrations. The results of our mathematical modelling will be verified with in vitro and in vivo experiments.